A Bayesian model to analyse the association of comorbidities with biosimilar treatment retention in a non-medical switch scenario in patients with inflammatory rheumatic musculoskeletal diseases

Abstract

Abstract Objectives To analyse clinical outcomes of a non-medical switch from originator adalimumab (ADA) to its ABP501 biosimilar (ABP) over 6 months in patients with inflammatory rheumatic musculoskeletal diseases (RMD) in relation to comorbidity as a risk factor for therapy discontinuation. Methods RMD patients switching from originator ADA to ABP were identified from a large routine database from October 2018 onwards. Documented clinical data at the time of non-medical switching (baseline), and at 3 and 6 months were collected. Comorbidities were represented by the Charlson Comorbidity Index (CCI) at baseline and patients were categorized based on CCI > 0. Differences in the ABP retention rate over 6 months between patients with CCI = 0 and patients with CCI > 0 were analysed using Bayesian exponential regression. Results A total of 111 patients with axial spondyloarthritis (n = 68), rheumatoid arthritis (n = 23) and psoriatic arthritis (n = 15), were identified, 74.8% of whom had continued treatment with ABP after 6 months, while a smaller proportion had either switched to another ADA biosimilar (10.8%), switched back to originator ADA (7.2%), switched to a different biologic (3.6%), or dropped out (3.6%). At baseline, a CCI > 0 was found in 38% of patients. Cardiovascular comorbidities (40%) were most prevalent followed by diseases of the skin (33%), the gastrointestinal tract (20%) and the eye (20%). ABP treatment was continued after 6 months in 74% of patients with CCI = 0 and in 76% with CCI > 0. Bayesian analysis showed only a small difference (months) in the APB continuation rate between groups (estimate 0.0012, 95% credible interval (CrI) -0.0337 to 0.0361). Adjusting for age, sex, and disease subtype revealed somewhat shorter retention rates for patients with CCI > 0, but the distribution of the difference included 0 (estimate -0.0689, 95% CrI -0.2246 to 0.0234). Conclusion In a non-medical switch scenario of RMD patients, there was no evidence for a considerable difference in ABP retention rates over 6 months between comorbidity groups.